Quantifying the impact of mitral valve anatomy on clinical markers using surrogate models and sensitivity analysis

Abstract

Blood flow studies within the left ventricle have proven to be promising for future clinical decision-making. However, accurate segmentation of heart valves, particularly the mitral valve, is still challenging. The MV has a significant impact on local flow phenomena within the ventricle and assumptions on its anatomy and location introduce uncertainties that are not yet fully understood. The overall aim of this study is to quantify the impact of uncertainty in defining MV anatomy and location on local and global clinical outcomes, such as kinetic energy, energy loss, transventricular pressure gradient and locally resolved wall shear stresses. We use a combination of computational fluid dynamics moving mesh simulations of cardiac blood flow, reduced order modeling (ROM) and variance-based global sensitivity analysis (GSA). We uncover a non-linear relationship between geometrical uncertainties and flow biomarkers with mitral valve size and angle as the most important parameters. Uncertainty quantification of echocardiography measurements reveals a standard deviation between 5-30% for the different output markers. We further outline the importance of robust ROM and GSA as model choice can drastically influence the results. Our entire pipeline is summarized in the open source tool SASQUATCH - a framework for sensitivity analysis and uncertainty quantification in cardiac hemodynamics.